Information Architecture for Irish Grocery Retailers using. Business Intelligence Tools

Information Architecture for Irish Grocery Retailers using Business Intelligence Tools Patrick Doran A dissertation submitted in partial fulfilment of the requirements of Dublin Institute of Technology for the degree of M.Sc. in Computing (Knowledge Management) September 2007

I certify that this dissertation which I now submit for examination for the award of MSc in Computing (Knowledge Management), is entirely my own work and has not been taken from the work of others save and to the extent that such work has been cited and acknowledged within the test of my work. This dissertation was prepared according to the regulations for postgraduate study of the Dublin Institute of Technology and has not been submitted in whole or part for an award in any other Institute or University. The work reported on in this dissertation conforms to the principles and requirements of the Institute s guidelines for ethics in research. Signed: Date: DD Month Year i

1 ABSTRACT Business Intelligence (BI) systems have a pivotal role to play in assisting retail management in the Irish grocery retail sector. The emergence of super chain stores and the increasing competition within this sector only increases their need for an effective and reliable Business Intelligence (BI) system. However despite this growing need Business Intelligence (BI) systems as they are currently deployed and used are not supplying the various levels of management with the necessary information they require to make effective retail management decisions. The information and reports received from Business Intelligence (BI) systems lack structure, are missing important pieces of information and do not provide a holistic view of the enterprise wide data. In particular the separation of data into separate pools prevents managers from getting the cross sectional view of the organisation necessary for effective decision making. The solution to the problem involves the use of information architectures. These information architectures allow the retail grocery organisations to examine their information sources, information flow processes and establish what management s information requirements are. The use of an information architecture therefore enables a company to establish what it s needs are from a Business Intelligence (BI) system independent of any particular technology. Therefore it can fit its Business Intelligence (BI) tools around its own unique processes and needs in order to successfully meet the needs of the various levels of management. This research presents the findings of a survey into the usage of Business Intelligence (BI) systems currently within the Irish grocery retail sector. The findings indicate that retail management are not getting the information and reports they require to make effective decisions. Building on these findings, and the learning gained from a literature view into the area, an information architecture was developed. The information architecture is presented from various different end user perspectives. This information architecture has a number of applications to grocery retailers such as providing a guide to retailers developing a Business Intelligence (BI) system from scratch or as a guide to retailers currently using Business Intelligence (BI) systems on how to maximise their return from it. Key words: Business Intelligence, information architecture, Irish grocery retailers, ii

ACKNOWLEDGEMENTS I would like to express my sincere thanks to my supervisor Brendan Tierney, whose valuable and enlightened insights and guidance was essential to the completion of this research project. I would also like to offer my sincere thanks to my wife Nuala and children Ava and Brian for all their support. iii

TABLE OF CONTENTS TABLE OF FIGURES... VII TABLE OF TABLES... VIIX 1. INTRODUCTION... 1 1.1 RESEARCH BACKGROUND... 2 1.2 RESEARCH PROBLEM... 3 1.3 RESEARCH OBJECTIVES... 5 1.4 INTELLECTUAL CHALLENGES... 6 1.5 RESEARCH METHODOLOGY... 7 1.6 RESOURCES... 8 1.7 PROJECT DELIVERABLES... 9 1.8 SCOPE AND LIMITATIONS... 9 1.9 ORGANISATION OF THE DISSERTATION... 10 2 BUSINESS INTELLIGENCE TOOLS... 13 2.1 INTRODUCTION... 13 2.2 DATA WAREHOUSING... 14 2.2.1 Advantages of data warehousing to retailers... 16 2.2.2 Conceptual design of a data warehouse... 18 2.2.3 Data warehouse architecture... 19 2.3 DATA MINING... 22 2.3.1 Data mining techniques... 24 2.2.6 Data mining techniques... 25 2.2.7 Retail companies using Data mining... 27 2.4 ONLINE ANALYTICAL PROCESSING (OLAP)... 28 2.5 GEOGRAPHIC INFORMATION SYSTEMS (GIS)... 29 2.5.1 GIS and OLAP... 30 2.6 VISUALISATION... 33 2.7 CONCLUSION... 34 3 ANALYSIS OF THE NEEDS OF RETAILERS DRIVING BUSINESS INTELLIGENCE... 36 iv

3.1 INTRODUCTION... 36 3.2 DRIVERS OF BUSINESS INTELLIGENCE (BI) SYSTEMS... 36 3.2.1 Customer Relationship Management the primary driver of Business Intelligence Systems.... 37 3.2.2 Business Performance Management (BPM)... 40 3.2.3 Emergence of superstores and increased competition.... 41 3.3 APPLICATIONS OF BUSINESS INTELLIGENCE TOOLS TO DECISION MAKING PROCESS IN IRISH GROCERY RETAILERS.... 42 3.3.1 Customer relationship management (CRM) applications... 43 3.4 RETAILERS REQUIREMENTS FROM AN INFORMATION ARCHITECTURE... 45 3.4.1 What is an information architecture?... 45 3.4.2. Information architecture evolution... 47 3.4.3 Developing an information architecture... 48 3.4.4 Understanding grocery retailers reporting requirements... 50 3.5 CONCLUSION... 51 4 ANALYSIS OF THE USGAE OF BUSINESS INTELLIGENECE (BI) TOOLS IN IRISH GROCERY REATILERS.... 53 4.1 INTRODUCTION... 53 4.2 RESPONDENTS TO THE SURVEY... 54 4.3 WHAT DO RETAIL MANAGERS NEED TO KNOW... 54 4.4 CURRENT LEVEL OF KNOWLEDGE OF RETAILERS... 56 4.5 RETAIL MANAGERS SATISFACTION WITH REPORTS RECEIVED... 59 4.6 PRESENTATION OF EXTRACTED INFORMATION AND REPORTS... 61 4.7 CAPABILITY OF EXTRACTING THEIR OWN REPORTS... 62 4.8 REPORTING IMPROVEMENTS... 63 4.9 KEY FINDINGS AND CONCLUSIONS... 64 5 INFORMATION ARCHITECTURE FOR IRISH GROCERY RETAILERS. 66 5.1 INTRODUCTION... 66 5.2 METHODOLOGY USED FOR CREATING THE INFORMATION ARCHITECTURE FOR IRISH GROCERY RETAILERS.... 66 5.3 INFORMATION ARCHITECTURE FOR STORE AND DEPARTMENT MANAGEMENT.. 68 5.4 INFORMATION ARCHITECTURE FOR REGIONAL AND SENIOR MANAGEMENT... 73 v

5.5 INTEGRATED INFORMATION ARCHITECTURE... 77 5.6 CONCLUSIONS... 80 6 EVALUATION OF THE INFORMATION ARCHITECTURE... 83 6.1 INTRODUCTION... 83 6.2 EXPERIMENTATION... 83 6.3 EVALUATION... 84 6.4 LEARNING GAINED FROM THE EVALUATION AND CONCLUSIONS... 87 7 CONCLUSION AND RECOMMENDATIONS... 89 7.1 INTRODUCTION... 89 7.2 RESEARCH DEFINITION & RESEARCH OVERVIEW... 89 7.3 CONTRIBUTIONS TO THE BODY OF KNOWLEDGE... 90 7.4 EXPERIMENTATION, EVALUATION AND LIMITATION... 92 7.5 FUTURE WORK & RESEARCH... 92 7.6 CONCLUSION... 93 BIBLIOGRAPHY... 95 APPENDIX A... 101 APPENDIX B... 107 APPENDIX C... 109 vi

TABLE OF FIGURES FIGURE 1.1 PROJECT PHASES (AUTHOR ((2007)).... 11 FIGURE 2.1 CONCEPTS OF DATA WAREHOUSING (LAUDON & LAUDON ((2005))... 18 FIGURE 2.2 EXAMPLE OF STAR SCHEMA FOR DATA WAREHOUSE (HAN & KAMBER (2006))... 20 FIGURE 2.3 EXAMPLE OF SNOWFLAKE SCHEMA FOR DATA WAREHOUSING (HAN & KAMBER (2006))... 21 FIGURE 2.4 EXAMPLE OF FACT CONSTELLATION SCHEMA FOR DATA WAREHOUSING (HAN & KAMBER (2006))... 22 FIGURE 2.5 STAGES IN THE DATA MINING PROCESS (CHAN ET AL., (2002)).... 26 FIGURE 2.6 EXAMPLE OF COMBINED GIS AND OLAP DATA (VOSS ET AL., ((2004))... 32 FIGURE 2.7 COMBINED GIS AND OLAP DATA (VOSS ET AL., ((2004))... 32 FIGURE 3.1 DRIVERS OF BUSINESS INTELLIGENCE (BI) SYSTEMS (AUTHOR (2007)).... 37 FIGURE 3.2 CRM APPLICATIONS (SEYBOLD GROUP (2002))... 39 FIGURE 3.3 APPLICATION AREAS OF BUSINESS INTELLIGENCE (BI) TOOLS FOR RETAIL ORGANISATIONS (AUTHOR, (2007)).... 43 FIGURE 3.4 EVOLUTION OF THE CONCEPT OF INFORMATION ARCHITECTURES (EVERDEN & EVERDEN, (2003)).... 48 FIGURE 3.5 DIFFERENT TYPES OF REPORTS REQUIRED BY DIFFERENT LEVEL OF MANAGEMENT AND THE TYPE OF DECISION THEY ARE USED TO ADDRESS (AUTHOR, ((2007).... 50 FIGURE 4.1 POSITION OF RESPONDENTS IN THE RETAIL ORGANISATION (AUTHOR, (2007)).... 54 FIGURE 4.2 RETAIL MANAGERS RATING OF THE INFORMATION AND REPORTS THEY RECEIVE (AUTHOR, (2007)).... 59 FIGURE 4.3 REGIONAL MANAGERS OPINION OF THE INFORMATION THEY RECEIVE (AUTHOR, (2007)).... 61 FIGURE 4.4 STORE MANAGERS AND DEPARTMENT MANAGERS OPINION OF THE INFORMATION THEY RECEIVE (AUTHOR, (2007)).... 61 FIGURE 4.5 MANAGERS RATING OF HOW WELL THE INFORMATION WAS PRESENTED IN REPORTS IN TERMS OF HOW EASY IT WAS TO UNDERSTAND (AUTHOR, (2007)).... 62 vii

FIGURE 4.6 NUMBER OF MANAGERS CAPABLE OF EXTRACTING THEIR OWN REPORTS (AUTHOR, (2007)).... 63 FIGURE 4.7 FIRST SUGGESTION MADE BY RETAIL MANAGERS FOR CHANGES TO THE REPORTS (AUTHOR, (2007)).... 64 FIGURE 5.1 FACTORS CONSIDERED DURING THE DEVELOPMENT OF AN INFORMATION ARCHITECTURE FOR GROCERY RETAILERS (AUTHOR, (2007)).... 67 FIGURE 5.2 INFORMATION ARCHITECTURE FOR STORE AND DEPARTMENT MANAGERS (AUTHOR, (2007)).... 69 FIGURE 5.3 INFORMATION ARCHITECTURE FOR REGIONAL AND SENIOR LEVEL MANAGEMENT (AUTHOR, (2007))... 74 FIGURE 5.4 INTEGRATED INFORMATION ARCHITECTURE (AUTHOR, (2007))... 79 viii

TABLE OF TABLES TABLE 4.1 RATINGS OF THE IMPORTANCE OF INDIVIDUAL PIECES OF INFORMATION TO THE RETAIL DECISION MAKING PROCESS (AUTHOR, (2007)).... 55 TABLE 4.2 RESPONSE TO QUESTIONS REGARDING THE RETAILER KNOWLEDGE OF CERTAIN FACTS (AUTHOR, (2007)).... 56 TABLE 4.3 WHEN THE ANSWER WAS YES IN TABLE 4.2 WHAT THE SOURCE OF THEIR KNOWLEDGE (AUTHOR, (2007))... 57 ix

1. INTRODUCTION The information economy puts a premium on high quality actionable information exactly what Business Intelligence (BI) tools like data warehousing, data mining, and OLAP can provide to the retailers. Haigang (2005). Dispersion of information sources and decentralisation of the decision making process has resulted in the insufficiency of present information management tools. In this situation, organisations are offered the BI systems applications Olszak & Ziemba (2006). The term Business Intelligence (BI) was coined by Howard Dresner of the Gartner Group in 1989 to describe the various concepts and methods available to enhance the business decision making process through the use of information systems (Hashmi, 2004). The information systems referred to included executive information systems, decision support systems, enterprise information systems, management information systems, online analytical processing (OLAP), data mining, geographical information systems (GIS) and data visualisation. Chou & Tripuramallu (2006) noted that some of these enterprise systems have reporting and basic query functions, however organisational data by its nature is scattered in various business information systems. Therefore the isolated reporting functions of the different systems is insufficient in providing a company with an organisational wide view of business operations. This necessity for an organisational wide view of business processes is one of the key drivers of Business Intelligence (BI) systems. Olszak & Ziemba (2006) presents Business Intelligence (BI) systems as providing the holistic infrastructure necessary to support decision making in organisations. The focus of this research project is on the applications of Business Intelligence (BI) tools to the Irish retail grocery sector. This sector of Irish retailing is very competitive and companies are looking towards Business Intelligence (BI) tools to provide them with competitive advantage over their market rivals (Chou & Tripuramallu, 2006) or perhaps it is now a competitive necessity. In contrast to the traditional reporting tools 1

such as spreadsheets, Business Intelligence (BI) systems are capable of providing a user friendly interface for examining multidimensional data sources. This provides the decision maker with access to actionable information in a timely manner thereby improving the decision making process. Business Intelligence (BI) tools can be used to answer various Customer Relationship Management (CRM) questions for example: (CRM2day.com, 2004) Who are your most valuable and least valuable customers? What factors affect the sale? How successful are promotional offers? Why are there variations between outlet s profits in different geographical locations? In addition to answering Customer relationship management (CRM) questions Business Intelligence (BI) tools also have various applications to Irish grocery retailers in other business functions (Haigang, 2005). For example Rasmussen et al., (2002) presents a case study of a company who were using Business Intelligence (BI) tools to extract actionable information from their various financial systems to answer questions such as: Sales order entry: sales by customer, sales by salesperson, sales by region, sales by outlet. General ledger: sales and profit by channel; actual various budgeted. Business Intelligence (BI) tools play a pivotal role in enhancing the quality of the decision making process facing Irish grocery retailers. In particular it provides an organisational wide view of the data and facilitates querying across different business functions. 1.1 Research background From the literature it is clear that there is no universal definition of Business Intelligence (BI) systems. However as observed by Negash (2004) they all refer to tools which are concerned with gathering and combining operational data with various analytical tools in order to provide managers with actionable information in a real time environment. Demand for BI technology is growing rapidly (Soejarto, 2003; Whiting, 2003). The use of the term Business Intelligence (BI) Systems refers to the set of related Business Intelligence (BI) tools working together to harvest information from 2

various different sources. The primary driver behind the increasing demand for Business Intelligence (BI) tools is the intense competition facing retailers. There are various applications of Business Intelligence (BI) tools to Irish grocery retailers particularly in the area of data warehousing and data mining. However there are no studies, to the knowledge of this author, concerned with assessing the relative success of these technologies in getting actionable information to retailers in the Irish context. These technologies must form part of an overall knowledge management system in order for the information to flow to all parts of the organisation. This point is particularly important with the advent of super chain-stores. In addition, retailers are faced with a bewildering array of Business Intelligence (BI) tools some of which they may not require or may not be suitable for their organisation. The author of this dissertation has a number of years experience in the application of data mining techniques and has worked with a number of large retailers in developing their retail information management systems. From this experience it has become apparent that there is huge confusion among Irish grocery retailers about the capabilities of the different Business Intelligence (BI) tools and how to apply them. Despite the widespread usage of these Business Intelligence (BI) tools many retailers struggle to get the actionable information to their retail managers in a timely fashion. In addition retail management are frustrated with the information contained within reports. There is a clear need for a new approach to the development and utilisation of Business Intelligence (BI) systems with the Irish grocery retail sector. 1.2 Research problem The aim of this research project was to create an information architecture, for Irish grocery retailers, for the use of Business Intelligence (BI) systems. Within the research literature there is no universally accepted definition of what an information architecture is. For the purposes of this research an information architecture is defined as a tool which provides an organisation with a mapping showing what data it captures, the location of this data and the uses and relationships between the data sources. It indicates to the Business Intelligence (BI) system where the data can be found and what data it needs to know in order to fulfil the information needs of the 3

various levels of retail management. This information architecture has a number of uses to Irish grocery retailers including the following: In the situation where grocery retailers are introducing Business Intelligence (BI) tools into a green field site it will provide a blueprint to be followed in order to provide the total solution. In this situation it will provide: I. A tool to establish what the information needs of the various levels of management are. II. A tool for identifying the scope of the project. It will allow the retailer to establish what Business Intelligence (BI) tools are most suitable to fulfil their information needs. III. A common language to facilitate collaboration concerning the project in particular between technical and non-technical personnel. IV. A tool to educate and train business users at the various different management levels of retail organisations. For grocery retailers who are already using Business Intelligence (BI) tools it will provide an architecture which they can compare to their own particular design. This will be beneficial to grocery retailers who are attempting to maximise the information return from their Business Intelligence (BI) systems. In this situation it will provide a tool for gap analysis. It allows grocery retailers to compare their design and layout of Business Intelligence (BI) tools against the architecture to identify gaps. For academic purposes it will be useful as a teaching tool and also a source of further research. The information architecture can be used to provide students with an overview of the different Business Intelligence (BI) tools available to grocery retailers and how these tools interact with each other. It will also be useful to identify areas which warrant further research. To create this information architecture it was necessary to perform a literature review into the area. This literature was broken into two areas; firstly a review was performed critiquing the various Business Intelligence (BI) tools available to grocery retailers, secondly a literature review was performed reviewing what the grocery retailer s needs are from Business Intelligence (BI) systems. Additionally in order to create the information architecture it was necessary to perform primary research in the form of a survey which was used to evaluate the current use of Business Intelligence (BI) tools 4

in Irish grocery retailers and determine what elements the different levels of retail management require in their reports. Babbie (1990) observed that the generic objective of all surveys used in research was to generalise from a sample to a population so that conclusions can be made. A survey was used for data collection because of the economy of design and the quick collection of data. The survey was cross sectional i.e. the data was collected at one point in time. The specific form of the survey was that of an administered questionnaire as described in the literature (Fink, 1995). The survey was administered over the telephone. The targeted population for the study was retail managers in Irish grocery retailers. These retail managers where divided into two groups, firstly store and department managers and secondly regional and senior level managers. The sampling design for this population was a single stage due to the fact that the author has access to names in the population and has access to them. The sample was selected for the study using the approach highlighted by Babbie (1990) known as a non probability sample whereby respondents are chosen based on their convenience and availability. In all an attempt was made to contact and survey 100 retail mangers with 66 of those contacted completing the survey. The survey data was analysed using Statistical Package for Social Science (SPSS) version 14.0. 1.3 Research objectives The following objectives have been achieved throughout the dissertation and contributed to the overall outcome: 1. A literature review was performed critiquing the various different Business Intelligence (BI) tools available to grocery retailers in particular their different applications. 2. A literature review was also performed to investigate and establish the needs of Irish grocery retailers from Business Intelligence tools. 3. An evaluation of the current information reports received by retail managers was performed through the primary research. The reports were analysed in terms of its quality, accuracy, timeliness and relevance to retail management decision making process. 5

4. The elements that the different level of retail managers require in their information reports was also established through the analysis of the survey data. 5. An information architecture was developed for Business Intelligence (BI) systems. This information architecture was developed from the analysis of the survey data and learning gained from performing the literature review into the area. 6. An evaluation of the information architecture was performed through a structured interview with relevant parties. 1.4 Intellectual challenges There were six intellectual challenges addressed in the process of completing this research. These challenges were addressed in the order below as the learning from one facilitated overcoming the next. Understand the capabilities, limitations and applications of the various different Business Intelligence (BI) tools currently available to Irish grocery retailers. Understand and identify the retailers need for Business Intelligence (BI) tools and their requirement from these tools. Understand how retailers are currently using Business Intelligence (BI) tools and how effective and appropriate the reports they extract from their systems are. Understand, establish and comprehend what the different levels of retail management reporting requirements are from Business Intelligence (BI) systems. Develop an information architecture for Business Intelligence (BI) systems then will enable them to effectively use their systems. Understand and integrate the feedback from the commercial personnel regarding the information architecture. 6

1.5 Research methodology Both primary and secondary research was used during the creation of this dissertation. The secondary research took the form of an extensive literature review into the area of Business Intelligence (BI) tools. The literature review was broken into two areas: Review of the latest trends and developments in Business Intelligence (BI) tools and their possible applications to Irish grocery retailers. Review into the needs of grocery retailers which is creating the demand for Business Intelligence (BI) tools and what their requirements are from these tools. Various different sources were used to complete the literature review including the following: Journals White papers Conference proceedings Newspapers Company websites. The primary research used took the form of a survey which was administered to retail managers. The survey was used to establish the following: Highlight the weakness and information gaps experienced by retailers using Business Intelligence (BI) tools and therefore the need for retailers to use an information architecture. Evaluate the reports currently received by retail managers. This evaluation was based on a number of criteria including the following: o Timeliness o Relevance o Accuracy o Understand ability Establish the elements that grocery retailers at different level of the management require in Business Intelligence (BI) reports. The knowledge gained through analysing the results of the surveys combined with that from the literature review was used to create the information architecture. A further element of primary research was then performed to evaluate the relevance and 7

accuracy of this information architecture to commercial organisations. This primary research was in the form of structured interviews. The information architecture was presented from different perspectives and also evaluated from these perspectives. A store manager, regional manager and information technology manager were therefore interviewed to establish their views regarding the information architecture. 1.6 Resources The availability of various resources was crucial to the completion of this dissertation. The following is a list of these resources: Library facilities Access to Dublin Institute of Technology s library facilities was essential for the completion of this dissertation particularly for the completion of the literature review. The ability to access library facilities from home also proved very useful. Computer with network access Availability of a computer with network access was essential as it provided access to various network printers and also provided a secure location to store the files. In addition to computer network access having the use of photocopying facilities was also useful. Microsoft Office Suit and Statistical Package for Social Science (SPSS). Having access to a computer which had the Microsoft office suite was crucial to the completion of the project. Also having access to Statistical Package for Social Science (SPSS) was important as it allowed for the analysis of the survey data. Both access to the Microsoft office suit and Statistical Package for Social Science (SPSS) was provided by the computer facilities in Dublin Institute of Technology. Internet access and email. The availability of internet access and usage of email facilitated the completion of this research project. Contacts with retail sector The author of this dissertation has a number of years experience working with different grocery retailers to improve the reporting capacity of this retail information management systems, in addition the author also has contact with a large number of retail managers through a continuous professional development degree programme in Dublin Institute of Technology. This proved very useful as 8

it provided easy access to a sample when performing the survey. Also it was useful because it meant the author had access to these individuals throughout the processes involved in this research. Guidance from supervisor Regular contact with the project supervisor was a crucial importance in completing the research. 1.7 Project deliverables The primary deliverable at the end of this research project is an information architecture for Business Intelligence (BI) tools. This information architecture is for companies in the Irish grocery retail sector. This research project also provides a set of guidelines to any retail company who are using Business Intelligence (BI) tools or considering introducing them to their organisation. The deliverables at the end of this project will consist of the following: Critical review of all available Business Intelligence (BI) tools available to Irish grocery retailers. Critical review of the retailers need for Business Intelligence (BI) and possible applications of Business Intelligence (BI). Report on how well retailers are currently using Business Intelligence (BI) tools with regard to the information reports they extract from their system. Report on what retail mangers require in the reports from their Business Intelligence (BI) systems. Description of information architecture showing each of the elements of the architecture and relationship between them. Commercial evaluation of the information architecture. 1.8 Scope and limitations The aim of this research project was to create an information architecture, for Irish grocery retailers, for the use of Business Intelligence (BI) tools. This information architecture will ensure that grocery retailers gain the maximum return from their Business Intelligence (BI) systems. This research focused on the grocery retail sector however there are number of other retail sectors that are also using Business 9

Intelligence (BI) tools that would also benefit from the development of an information architecture. This is however beyond the scope of this research. The author intends to use this information architecture in his work with grocery retailers, however to assess its usefulness would require longitudinal research which is beyond the scope of this research project. The evaluation method involved presenting the information architecture to a relevant party and performing a structured interview. However to thoroughly evaluate the information architecture it would be necessary to introduce it into a retailer organisation and then evaluate its effects of the information extracted from the system. 1.9 Organisation of the dissertation The major phases of the research project are shown in figure 1.1 and are directly related to the chapters in the dissertation. Chapter 2 of the dissertation presents the findings of the literature review into the different Business Intelligence (BI) tools available to Irish grocery retailers. This chapter focuses on the latest developments in each of the Business Intelligence (BI) tools considered in this research project and their applications to grocery retailers. The Business Intelligence (BI) tools reviewed are; Data warehousing Data mining Online analytical processing (OLAP) Geographic information systems (GIS) Visualisation technologies. Chapter 3 of the dissertation presents the findings of the literature review into the needs of grocery retailers from Business Intelligence (BI) systems. It begins by discussing the various different drivers of Business Intelligence (BI) systems. It details each of the drivers individually and also describes the major overlap between the drivers. It then describes the different applications of Business Intelligence (BI) systems to Irish grocery retailers which focuses primarily on the customer relationship management (CRM) applications. The final section of this chapter details the concept of information architectures. This section begins by discussing the various different definitions of information architecture and ends presenting the definition used in this research. 10

Critical review of BI tools Review of Business Intelligence (BI) tools Critical review of Retailers Needs Review of Retailers need for Business Intelligence (BI) Retailers Perform Survey Analysis of how successfully retailers are at using BI tools Report on use of Business Intelligence (BI) tools Analysis of what retail managers require from BI Commercial appraisal Information architecture Conclusion Process order Output Information flow Figure 1.1 Project phases (Author ((2007)). 11

Chapter 4 of the dissertation contains the analysis of the survey data into the usage of Business Intelligence (BI) tools in Irish grocery retailers. This chapter presents the findings to the various questions used in the survey. These findings provided the knowledge required to build the information architecture. Chapter 5 presents the information architecture. This chapter is broken into a number of sections each of which present the information architecture from a particular perspective. The information architecture is presented from three different perspectives; store and department manager s perspective, regional and senior management s perspective and the integrated perspective which is intended for information technology personnel or technology confident business people. Chapter 6 contains the evaluation of the information architecture. This chapter contains the findings of the structured interviews carried out with the relevant parties. The responses to the questions are summarised and noted in this chapter and a discussion of the findings are presented. Finally conclusions regarding the usefulness and accuracy of the information architecture are presented and discussed. Chapter 7 is the final chapter of the dissertation and it contains the conclusions and recommendations formed after the completion of this research project. 12

2 BUSINESS INTELLIGENCE TOOLS 2.1 Introduction Thomsen (2003) noted that the term Business Intelligence (BI) system has replaced decision support, executive information systems and management information systems. With each evolutionary step the systems have improved in terms of their capability to meet the enterprise s need for computational and analytical queries (Power, 2004). Negash (2004) presents the following definition for Business Intelligence (BI): BI systems combine data gathering, data storage, and knowledge management with analytical tools to present complex internal and competitive information to planners and decision makers. Fundamental to the idea of Business Intelligence (BI) is the concept of getting actionable information to managers in a timely fashion. Yet despite this clearly stated objective there is complete lack of research into the area of assessing how successful these systems have been at achieving this (Negash, 2004). Langseth & Vivatrat (2003) produced a list of the essential components of proactive Business Intelligence (BI) systems which includes the following; Data warehousing, which is performed in real time Data mining capabilities Automatic detection of anomalies and exceptions Seamless follow-through workflow Geographic information systems (GIS) Data visualisation. Business Intelligence systems play a pivotal role in assisting managers with strategic and operational decision making. Willen (2002) quoted a Gartner survey which ranked the strategic use of Business Intelligence systems as follows: (1) Corporate performance management 13

(2) Improving and maximising customer relationship management (CRM), monitoring business activity and providing the data necessary to support decision making (3) Specific Business Intelligence (BI) applications for targeted operations (4) Management reporting of Business Intelligence (BI). The primary objective of using Business Intelligence (BI) systems in Irish grocery retailers is to support the making of various decisions on production, sales, competition monitoring and finance (Kalkaota & Robinson, 1999). 2.2 Data warehousing Bill Inmon coined the phrase data warehouse in 1992 and defined it as a managed database in which the data is: 1. Subject oriented: as distinct from application-oriented, i.e. data designed to aid in decision making. If a data warehouse is well designed it will provides a stable image of business processes, independent of legacy systems. For example it could be based around customers. 2. Integrated: the data warehouse consolidates application data from different legacy systems, usually operational databases which may use different coding and measurement units therefore cleansing and integrating is a key task. The pivotal operational system feeding the data warehouse for Irish grocery retailers is the electronic point of sales systems (epos). Another example of the sources of data which maybe integrated into the data warehouse includes data from the loyalty card system. 3. Time-variant: all information in the data warehouse will have a time dimension i.e. each data point is associated with a point in time, and data points can be compared along that time axis unlike operational databases which only content up to date information. This inclusion of historical data provides the retailer with the ability to compare sales through time. 4. Non-volatile: old data is not replaced instead new data is always appended. The data warehouse will absorb new data, integrating it with the previous data (Inmon, 1992). 14

The primary reason that Irish grocery retailers have and are using data warehouses is to build a database (data warehouse) separately from transactional databases. This is due to the fact that analytic data and transactional data are different in terms of requirements and user communities. The data warehouse mainly stores detailed summarised data and metadata. Data within the data warehouse is typically aggregated to improve effectiveness of queries e.g. sales maybe aggregated by geographical dimensions or by time. The metadata contained within the data warehouse is concerned with facilitating the process of extracting, transforming and loading data from the various operational sources into the data warehouse (Forcht, 1999). Metadata is also concerned with automating the summarisation of data query management. The data warehouse provides an integrated data repository to assist retail managers in the decision making process. In order for the data within the data warehouse to be of high quality the data from the operational databases must be effectively and efficiently placed into the data warehouse. ETL (Extraction-Transformation-Load) tools are concerned with getting the data from the various operational sources into the data warehouse (Olszak & Ziemba, 2006). Meyer (2001) divided ETL tools into four categories as follows: ETL tools that prefer specific types of input or output data and are reliable with fast functions of data processing and transforming; ETL tools that focus on extraction and loading of data; ETL tools that perform the process of data transformation quite well, although they do not offer servicing of many data formats; ETL tools that provide complex integration environments equipped with numerous solutions to assist users while constructing ETL systems. Extraction of the data begins with getting access to the original data which is typically stored in various operational relational databases. The data from different operational sources will usually have differing formats and this greatly increases the complexity of the process. After extraction data is typically stored in a relational database which allows further data processing, this is known as a staging area. The extraction time, structure of source data and other details are typically recorded by the software which 15

extracts the data (Olszak& Ziemba, 2006). The next stage, transformation, is the most complex. During transformation a scripting language, typically SQL, performs data unification, aggregation calculation and identification of missing data. These processes will be performed according to a set of rules. The final stage involves loading the aggregated and filtered data into the data warehouse. Each retail organisation must decide upon the appropriate level of aggregation which meets their particular needs. It is important that retail management are responsible for making the decisions regarding the appropriate level of aggregation. 2.2.1 Advantages of data warehousing to retailers The research literature presents various advantages of data warehousing which are summarized in the following section (Zeng et al., 2003; Lawyer et al., 2004). 1. Simplicity Data warehousing simplifies business decision making by providing a single repository of data from which to generate a picture of the business. This picture of the business will include data from various sources that have been integrated into the data warehouse. Current operations can be monitored and compared with past operations, predictions of future operations can be rationally made and new business processes can be devised. Data warehouses typically store large amounts of historical data and corporate-wide data which companies need to turn into vital business information which they can base their decisions upon. 2. Better quality data. Data warehousing improves the information that the various levels of managers receive in terms of consistency, accuracy, and documentation. The fact that the data is extracted transformed and loaded from the operational systems means that it should of high quality and provide a cross functional view of the organisation. The holistic view of data greatly improves the decision making ability of management. 3. Fast access. Data warehouses allow users to retrieve necessary data by themselves in a timely fashion. Correctly designed and installed data warehouses are the foundation on which Business Intelligence (BI) systems are built. These 16

systems should allow the user to access up to date information though a user friendly interface. 4. Easy to use. Queries from users do not interfere with operational systems, because a data warehouse enables easy access to business data without slowing down the operational database by taking operational data, aggregating it, and loading it into a separate data warehouse. The data warehouses focuses on subjects which conceptually makes it easier for retail mangers to understand. The data warehouse should have a user friendly interface which permits the business user to easily query the data warehouse. 5. Separation of decision-support operation from production operation. Another advantage is that data warehouses are built in order to separate operational, continually updated transaction data from historical, more static data required for business analysis. By doing so, managers and analysts can use historical data for their decision-making activities without slowing down the operational systems. This separation also allows retail managers to store the historical data which is essential in retail decision making. 6. Source of competitive advantage. By using a data warehouse and thereby improving the management and utilisation of corporate knowledge, companies become more competitive, better understand customers, and more rapidly respond to dynamic customers. Therefore it could be argued that it is a source of competitive advantage. However the author of this dissertation is of the opinion that data warehousing similar to other Business Intelligence (BI) tools has now became a competitive necessity for Irish grocery retailers. 7. Information flow management. The usage of data warehouses improves a retailer s information management system. The data warehouses handles the large volumes of data from various operational data sources, and in so doing it manages the flow of information in addition to correlating the data. The use of a data warehouse ensures that all the data required by management in order to facilitate effective decision making can be located in one repository. 17

8. Security. The users of data warehouses do not have direct access to the operational systems of the grocery retailer. This ensures the security of the operational systems is not compromised. 2.2.2 Conceptual design of a data warehouse The primary concept of data warehousing is that the data stored for business analysis can most effectively be accessed by separating it from the data in the operational systems as displayed in figure 2.1. Figure 2.1 Concepts of data warehousing (Laudon & Laudon ((2005)). For data warehousing systems to be effective, as the foundation for Business Intelligence (BI) systems, data from various operational systems must be integrated into the data warehouse. In practice when integrating data from various sources it is easier to carry out the integration independent of the source applications. For grocery retailers the data warehouse must combine data from multiple source applications such as sales, marketing, finance, production and customer loyalty card system. Many large data warehouse architectures allow for the source applications to be integrated into the data warehouse incrementally (Chaudhuri et al., 1997). 18

The primary business reason for combining data from multiple source applications is to provide the capacity to cross-reference data from various business functions. All the data contained in a data warehouse should have a time tag. This is essential because data is never replaced in a data warehouse instead new data is appended. The data warehouse system can serve not only as an effective platform to merge data from multiple current applications; it can also integrate multiple versions of the same application. The data warehouse enables the retailer to perform year-on-year analysis even though the operational data concerned from historical years will no longer be contained in the operational systems (Han & Kamber, 2006). The most important reason for separating data for business decision making from the operational data is the possibility of performance degradation on the operational system that can result from the analysis processes. The execution of complex queries could slow down the performance of operational databases (Han & Kamber, 2006). This is of particular importance in the retail context as without their primary operational systems, electronic of sales systems (epos), they could not perform transactions. The fact that the data in a data warehouse is non-volatile is another important characteristic. In effect this means that after the data has been loaded into the data warehouse, no modifications can be made to it. In summary, the primary concept behind data warehousing is the separation of operational data from the data used for decision making. The operational data is aggregated and then loaded into a data warehouse which is completely separate from the operational databases. There are a number of business and technical issues that must be overcome in the data warehousing process. 2.2.3 Data warehouse architecture From the literature there are two main types of architectures which may be chosen when designing a data warehouse. The first architecture is known as the Bill Inmon architecture (Inmon et al., 1998) and the second is known as the Ralph Kimball architecture (Kimball, 1996). Both architectures are similar in that they both secure raw data from legacy batch systems in addition to online operational systems and specialised operational data stores (ODS). However they fundamentally differ in how 19

they structure the data within the data warehouse. The Inmon architecture uses an atomic-level, third-normal form (3NF) relational structure in contrast to the Kimball architecture which uses a multidimensional structure. The most popular model for data warehouses is the multidimensional model which can exist in three different forms star schema, snowflake schema or a fact constellation schema (Armstrong, 2002; Chen, 1976; Kronenke, 2002; McFadden, 1999). The most used architecture is the star configuration. An example of a star schema for a data warehouse in a retail organisation is shown in figure 2.2. The example is adapted from Han & Kamber (2006). There are two characteristics associated with the star schema for data warehouses: 1. They contain a central table, called a fact table, which contains the majority of the data, with no redundancy 2. They contain a number of smaller tables, called dimension tables that contain the data pertaining to each dimension. The arrangement of the tables result in it resembling a star, hence the name star schema. The example shown in figure 2.2 for a retail organisation show sales with four associated dimensions. The dimensions used in this example are time, branch, item and location. Figure 2.2 Example of Star schema for data warehouse (Han & Kamber (2006)). 20

The snowflake schema has many characteristics in common with the star schema. With the snowflake schema some of the dimensions tables are normalised resulting in the splitting of these tables into further tables. This results in a structure which resembles a snowflake as can be shown in the example of a snowflake schema shown in figure 2.3. Snowflake schema differs from the star schema in that the dimension tables maybe be all normalised to reduce redundancies. Figure 2.3 Example of snowflake schema for data warehousing (Han & Kamber (2006)). The fact constellation schema uses multiple fact tables to share dimensions tables. In reality is in truth a collection of star schemas hence it is sometimes called a galaxy schema. An example of the fact constellation schema is given in figure 2.4. The designing, creating and implementing of a data warehouse is a complex procedure, consisting of the following activities as outlined by Chaudhuri, (1997): Decide upon the architecture of the data warehouse, select the storage servers, database and online analytical processing (OLAP) servers, and tools. Integration of the servers, storage medians and client tools. Decide upon and design the warehouse schema and appropriate views. 21

Define the physical warehouse organization, data placement, partitioning, and access methods. Connect the data sources using gateways, ODBC drivers, or other wrappers. Implementation of suitable designed scripts for data extraction, cleaning, transformation, load, and refresh. Populate the data repository with the schema and view definitions, scripts, and the necessary metadata. Decide what end-user applications to use, design and implement them. Roll out the warehouse and its applications. However the approach outlined by Chaudhuri (1997) focuses exclusively on the technical issues and does take into account a number of business issues that must be addressed in developing a data warehouse. Figure 2.4 Example of fact constellation schema for data warehousing (Han & Kamber (2006)). 2.3 Data mining There are a number of different definitions of data mining in the literature (Ahmed, 2004; Fayyad and Uthurusamy, 2002; Hauke et al., 2003; Kantardzic, 2002; Poul et al., 2003). All definitions regarding data mining emphasise that it is concerned with 22

discovering various patterns, making generalisations or finding rules to describe data sources. Fayyad et al. (2002) defined data mining as;..the identification of interesting structure in data. Structure designates patterns, statistical or predictive models of the data, and relationships among parts of the data.. Each of the terms in the definitions, patterns, models and relationships, have concrete definitions. The term pattern refers to an identifiable trend or feature in the data or a subset of the data. A model is used to describe the data and anticipate future consumer behaviour. The final term relationship is concerned with the fact that there are connections and correlations between different pieces of data. The three main uses of data mining to retailers are classified as follows (Haigang, 2005); Providing information to enhance the customer relationship management (CRM) functions in particular in the area of customer attraction and retention. If grocery retailers are to perform effective customer relationship management (CRM) they must be capable of extracting and understanding information and their customers. Identification of fraud. The use of data mining enables managers to identify any unusual trends which could possibly be an indicator of fraudulent activities. Identification of operational inefficiencies within and between retail outlets. The use of data mining provides the retailer with capacity to identify operational inefficiencies within the outlet in addition to comparing the performance of similar outlets in different locations. This can identify areas which senior must address. Generally the information discovered from data mining is concerned with prediction or description of reality. Prediction endeavours to use historical data to forecast future customer behaviour whereas description attempts to describe and explain the action of customers. For example data mining customer s purchases details can form the basis on a decision regarding future pricing policy (Moss & Alert, 2003; Reinschmidt & Francoise, 2001). 23